BU Division of Systems Engineering

PowerPoint Presentation: 

WELCOME TO THE DIVISION OF SYSTEMS ENGINEERING Ruth Mason Division Director Elizabeth Flagg, Ed.M. Graduate Programs Coordinator Christos G. Cassandras SE Division Head Pirooz Vakili SE Associate Division Head Yannis Paschalidis CISE Co-Director David Castanon CISE Co-Director Linda Grosser CISE - Associate Director Division - Director of External Relations SE DIVISION CENTER FOR INFORMATION AND SYSTEMS ENGINEERING (CISE) DIVISION OF SYSTEMS ENGINEERING

PowerPoint Presentation: 

D IVISION OF S YSTEMS E NGINEERING Unique interdisciplinary graduate program combining Information, Decision, and Control Sciences Select faculty from different departments in the College of Engineering + affiliated faculty from College of Arts and Sciences and School of Management Cutting-edge research through the Center for Information and Systems Engineering (CISE)

PowerPoint Presentation: 

DIVISION DIVISION OF SYSTEMS ENGINEERING DEPARTMENT 1 DEPARTMENT 2 DEPARTMENT 3 COLLEGE OF ENGINEERING … Graduate program only Emphasis on PhD level research WHAT’S A “DIVISION” ?

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING WHAT IS SYSTEMS ENGINEERING ? Systems Engineering transcends the physical nature of what is designed or managed Systems Engineering cuts across traditional engineering departments and “enables” building, analyzing, or managing a system – be it electrical, mechanical, chemical, biological, or business Graduates of the SE Division adapt their knowledge and expertise to different application domains. They easily adjust from one environment to another and apply their methodological tools to whatever the new “system” may be Our students are trained to communicate research results through poster competitions and frequent presentations at leading international scientific conferences

PowerPoint Presentation: 

Develop a mathematical/computational model , possibly including randomness Understand dynamics : if/how one part of the systems affects others Use feedback principles: control system behavior Measure performance in physical, engineering, and economic terms Try to improve and optimize performance DIVISION OF SYSTEMS ENGINEERING A “SYSTEMS APPROACH” RECIPE

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING STUDENT OPPORTUNITIES The Automated Design and Manufacturing Systems (ADMS) Laboratory Internships and Project Sponsors • ExxonMobil Research and Engineering • Siemens, Corporate Research Lab • The MathWorks • Lincoln Labs • SAP Labs, Palo Alto • Ember Corporation • Schlumberger • The Hartford Insurance • Microsoft Corp.

PowerPoint Presentation: 

DIVISION FACULTY ELECTRICAL AND COMPUTER ENGINEERING Christos Cassandras David Castanon Ioannis Paschalidis Venkatesh Saligrama David Starobinski Affiliated : Murat Alanyali Prakash Ishwar Clement Karl Lev Levitin Thomas Little Ari Trachtenberg MECHANICAL ENGINEERING Sean Anderson John Bailleul Calin Belta Michael Caramanis James Perkins Pirooz Vakili Hua Wang Affiliated : Michael Gevelber BIOMEDICAL ENGINEERING Affiliated : James Collins Sandor Vajda COMPUTER SCIENCE Affiliated : Azer Bestavros Mark Crovella Abraham Matta MATHEMATICS Affiliated : Eric Kolaczyk OPERATIONS AND TECHNOLOGY MANAGEMENT Affiliated : Erol Pekoz College of Engineering College of Arts and Sciences School of Management DIVISION OF SYSTEMS ENGINEERING

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING CURRICULUM The SE program leads to post-Bachelor’s Ph.D. and Post-Master’s Ph.D. degrees. M.S. and M.Eng. degree programs provide concentrations in - Computational and Systems Biology - Control Systems - Network Systems - Financial Engineering Systems - Manufacturing Systems and Supply Chains - Operations Research Currently 26 Ph.D. students Cooperative systems and sensor networks are studied in the Intelligent Mechatronics Lab and the Control of Discrete Event Systems Lab using small wireless robots

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING CORE COURSES AND COURSES BY CONCENTRATION AREA CORE SE/EC/ME501 Dynamic Systems Theory SE/EC/ME710 Dynamic Programming and Stochastic Control SE/EC524 Optimization Theory and Methods SE/ME714 Advanced Stochastic Modeling and Simulation EC505 Stochastic Processes EK500 Probability with Statistical Applications Computational and Systems Biology Operations Research Network Systems Control Systems Financial Engineering Systems Manufacturing Systems and Supply Chains

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING RESEARCH Automation, robotics, and control Communications and networking Systems biology Information sciences Production, service systems, and supply chains Research Funding ~ $3.3 M/year Sample projects Wireless sensor network test bed used for the Boston University Statistical Localization System (BLoc) A Control and Optimization Science Base for Sensor Networks in Adverse and Stochastic Environments (NSF) Communicating Networked Control Systems (ARO) Networked Sensing Systems for Urban Target Recognition (ONR) Distributed Wireless Sensor Networks for Long-term Deployments (DOE) Event-driven Sensing For Enterprise Reconfigurability and Optimization (NSF) Localized Computation and Network Path Formation to Enable Pervasive Video Sensing (NSF) Image-Guided Intracardiac Beating Heart Surgery (Children’s Hospital) Rational Design of Synthetic Gene Networks using Formal Analysis of Hybrid Systems (NSF) Final-Stage Optimization Methods for Protein Docking Exploiting Energy Funnels (NIH)

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING Field reconnaissance: Map level sets Map steepest ascent/descent curves SEARCH & RECONNAISSANCE - Intelligent Mechatronics Lab Sean Andersson Hua O. Wang John Baillieul Intelligent Mechatronics Lab Faculty

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING In exploration of an unknown random field , we seek to understand: What are the criteria for deciding whether information being gathered is interesting? What are the criteria for deciding what to explore next? A potential field is most interesting when its variability is most unpredictable. A field is a communication channel that provides information in the range about the nature of something in the domain X . One goal of a good search or reconnaissance strategy is to as rapidly as possible increase the value of an information metric like SEARCH & RECONNAISSANCE - Intelligent Mechatronics Lab

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING HYBRID AND NETWORKED SYSTEMS - Belta Verification and control of hybrid systems Formal approaches to planning and control of robot motion Modeling, analysis, and control of gene and metabolic networks C. Belta hyness.bu.edu

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING POWER MARKET RE-DESIGN - Caramanis Wholesale Power Market re-design (Smart Grid, uncertainty impact of intermittent generation and distributed resources) Reserve Contingency Planning ought to be Endogenous to Day Ahead – unit commitment – markets Sequential Markets – Day Ahead, Adjustment, Real time Introduction of Local Retail Markets Interacting with Wholesale Market Building/building-cluster micro grids for (i) monitoring and control of diverse loads, (ii) local reliability and self sufficiency, (iii) Effective system interaction Time-varying local/global cost reflective electricity rates: local congestion, losses, voltage support, wholesale market energy & reserve costs New Decision Support Algorithms: Network Science, DP, Robust Optim. Market Clearing must reflect (i) new uncertainties (ii) new participants on the load side, (iii) Interaction of sequential markets Individual Market Participant Actions: bids/offers to wholesale market associated with load management and response to local costs/congestion

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING COOPERATIVE CONTROL AND OPTIMIZATION – Cassandras 1. Distributed Asynchronous Optimization - Event-driven control and optimization (no synchronization) - Handle (and exploit…) uncertainty: Hedge-and-React control - Limit communication, without affecting optimal cooperation (energy, bandwidth constraints; security concerns) 2. Sensor Networks - Optimal deployment, persistent surveillance - Energy management – new battery models viewed as dynamic systems - Max. lifetime routing 3. Applications in the “city enterprise” - Intelligent parking - Deployment in environments with obstacles (Funding: NSF, AFOSR, ONR, DOE, MathWorks, Honeywell)

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING STOCHASTIC OPTIMIZATION, SIMULATION TOOLS – Cassandras 1. Optimization of stochastic networks - Stochastic Flow Models (SFMs): flow rates are random processes - Sample path gradient estimation - Resource contention games - No stochastic modeling assumptions required 2. Optimization in Probability (as opposed to conventional Optimization in Expectation) - Optimize over sample paths, then Estimate solution distribution - Solution optimal more frequently than any other solution - Reduce computational complexity 3. Discrete-Event and Hybrid System Modeling and Simulation - Modeling with SimEvents ® (The MathWorks) - Stochastic Optimization “toolbox” using SimEvents ® , Simulink ® , MATLAB ® Dynamic Voltage Scaling Controller

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING ESTIMATION AND STOCHASTIC CONTROL - Castañón Subsurface inverse problems NSF Engineering Research Center (CenSSIS) – medical and environmental applications Department of Homeland Security (DHS) center (ALERT) – luggage inspection systems, whole body imaging using multispectral, sparse reconstruction Adaptive detection systems/machine learning NSF/NIH project on machine learning for cancer detection using elastic spectroscopy DHS ALERT work on distributed mobile detection systems Stochastic Control for sensor management AFOSR + MURI sponsored work on scalable stochastic dynamic programming techniques for sensor management and trajectory control Adaptive mission control for unmanned air vehicles AFOSR – Adaptive mission control algorithms for increased autonomy in unmanned vehicles

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING TOPICS IN WIRELESS SENSOR NETWORKS (WSNETs) - Paschalidis Localization Node localization Source localization (e.g., in chem/bio detection) Probabilistic approach based on hypothesis testing Optimal deployment Testbed evaluation in warehouse management Energy management Energy-aware routing Media Access Control: Transmission scheduling Topology optimization for efficient in-network processing (consensus/averaging) Robust maximum lifetime routing Anomaly detection Statistical anomaly detection in the environment and/or the operation of the WSNET

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING PROTEIN DOCKING - Paschalidis What is protein docking ? A key problem in computational biology, elucidating protein-protein interactions Given the 3-D structure of two component proteins determine the 3-D structure of the complex (<5Å RMSD) Approach Gibbs free energy minimization (over SE(3)) Developed SDU (Semi-Definite Underestimation): a stochastic global optimization algorithm for funnel-like functions Results Success in CAPRI (community wide experiment/competition) Analysis of the protein docking landscape

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING (Funding: NSF, NGA(NURI), ONR, DHS) Learning Theory, High-Dimensional Stats Anomaly Detection in HighD spaces Uniformly Most Powerful; Computational Comp: Linear in Dim. Active Learning theory Sample complexity issues Video Analysis over Multi-Camera Networks Issue: Urban Scenarios --- Too many objects and Tracks Key Insights: Pixel level activity patterns --- suff. info for det. motion anomalies, invariant to zoom and view angles Compressive Sensing Filtered Sparse Processes (Neuronal Spikes, Blind De-convolution, …) Distributed Architectures Group Testing Graph Constrained Testing (Network Tomography) Boolean Compressed Sensing/Noisy Group Testing (Cognitive Radios) In-Network Signal Processing (Sensor Networks) Detection, Estimation, Tracking Message Passing Algorithms Packet losses, Energy Efficiencies, … STATISTICAL SIGNAL PROCESSING - Saligrama

PowerPoint Presentation: 

DIVISION OF SYSTEMS ENGINEERING EFFICIENT SIMULATION AND MONTE CARLO INTERPOLATION – Vakili 1. Parametric Variance Reduction (VR) & Monte Carlo Interpolation Generic approach to VR A learning framework for VR Flexible interpolation of computational information VR for Sensitivity estimation (IPA, LR, Malliavin) Finance, Physics, OR applications 2. Financing & Contracting in Renewable Energy Analysis of current financing & contracts Alternative contracts to promote adoption Pooling & bundling of services 3. Protein Docking Paschalidis’s slide